Alternative way to get genomic positions

[grst]

Description of feature

Currently the only way to get genomic positions is through reading a GTF file. This is (a) slow and (b) gtfparse repeatedly makes problems.

It could be more conveniente to retrieve this information from online sources such as biomart or Bioconductor AnnotationHub.

Then gtfparse could become an optional dependency.

[zktuong]

This works very well for me as well
https://scanpy.readthedocs.io/en/stable/generated/scanpy.queries.biomart_annotations.html

def query_biomart() -> pd.DataFrame:
    """
    Extract gene annotations from Biomart.

    Parameters
    ----------
    index_key : str, optional
        Index key for the DataFrame.

    Returns
    -------
    pd.DataFrame
        DataFrame with gene annotations from Biomart.
    """
    annot = sc.queries.biomart_annotations(
        "hsapiens",
        [
            "ensembl_gene_id",
            "hgnc_symbol",
            "start_position",
            "end_position",
            "chromosome_name",
        ],
        use_cache=True,
    ).rename(
        columns={
            "ensembl_gene_id": "gene_ids",
            "hgnc_symbol": "gene_symbol",
            "start_position": "start",
            "end_position": "end",
            "chromosome_name": "chromosome",
        }
    )
    return annot
    
def annotate_var(
    adata: AnnData, annotation: pd.DataFrame, index_key: str = "gene_ids"
) -> None:
    """
    Annotate the features with in an AnnData object.

    Parameters
    ----------
    adata : AnnData
        Input AnnData object.
    annotation : pd.DataFrame
        Gene annotation DataFrame.
    index_key : str, optional
        Index key for the DataFrame.
    """
    for col in ["start", "end", "chromosome", index_key]:
        assert (
            col in annotation.columns
        ), f"Annotation DataFrame must contain the column named `{col}`."

    for col in annotation:
        var_dict = annotation[col].to_dict()
        adata.var[col] = [
            var_dict[x] if x in var_dict else None for x in adata.var[index_key]
        ]

[grst]

very nice 🤩